Attempted solution for https://github.com/slowkow/ggrepel/issues/163

badrepel.R

library(ggplot2)
library(ggrepel)
library(dplyr)
library(stringr)
library(scales)
## https://github.com/tidyverse/ggplot2/issues/980
mysqrt_trans <- function() {
    trans_new("mysqrt",
              transform = base::sqrt,
              inverse = function(x) ifelse(x<0, 0, x^2),
              domain = c(0, Inf))
}

pd <- read.table(textConnection("
           MAST Wilcoxon           mtd
alra        0.8      0.5          ALRA
autoimpute  0.0      0.0    AutoImpute
baynorm     0.4      0.0       bayNorm
dca         0.0      0.0           DCA
deepimpute  7.4      0.0    DeepImpute
drimpute    0.2      0.0      DrImpute
knnsmooth   0.0      0.0 kNN-smoothing
magic       0.0      0.0         MAGIC
mcimpute    0.1      0.0      mcImpute
pblr        1.4      0.0          PBLR
raw         0.0      0.5        no_imp
saver       0.0      0.0         SAVER
saverx      0.0      0.0        SAVERX
scimpute    0.6      0.0      scImpute
screcover   0.0      0.1     scRecover
scscope     0.1      1.1       scScope
scVI        0.0      0.0          scVI
viper       0.3      0.0         VIPER"))

## Initial attempt
ggplot(pd) +
    aes(x = MAST, y = Wilcoxon, label = mtd, color = mtd) +
    geom_point() +
    geom_text_repel() +
    theme_bw() +
    theme(legend.position = 'none')

mult <- 0.1
pd_grouped <- pd %>%
    group_by(point=paste0(MAST, ",", Wilcoxon)) %>%
    summarise(MAST = MAST[1], Wilcoxon = Wilcoxon[1],
              mtd = paste(mtd, collapse="\n")) %>%
    mutate(xnudge = (max(MAST) - MAST) * mult,
           ynudge = (max(Wilcoxon) - Wilcoxon) * mult)
ggplot(pd_grouped) +
    aes(x = MAST, y = Wilcoxon, label = mtd, color = mtd) +
    geom_point() +
    geom_text_repel(nudge_x = pd_grouped$xnudge,
                    nudge_y = pd_grouped$ynudge) +
    theme_bw() +
    theme(legend.position = 'none') +
    scale_x_continuous(trans="mysqrt") +
    scale_y_continuous(trans="mysqrt")